微機械麥克風之最佳化設計及其陣列於3D聲場重建之實現

Abstract

電容式麥克風有較高的靈敏度與較平坦與寬廣的頻率響應，因此被廣泛的應用於各種場合。利用微機電技術可將體積微小化，卻不失其性能，並可降低雜散電容的影響。文中探討微機械電容式麥克風的動態行為分析，並利用田口實驗法(Taguchi method)與遺傳演算法(Genetic Algorithm)最佳化其性能與尺寸。然後利用最佳化之尺寸來設計微機械電容式麥克風，使其有最佳的靈敏度與頻寬。利用陣列訊號處理技術可提高陣列麥克風的訊噪比(Signal-to-Noise Ratio)與指向性(Directivity)。其中延遲補償法(Delay-Sum method)與超指向性法(Superdirective method)可用來做聲源方向估計(Direction of Arrival)與聲束形成(Beamforming)。設計超指向性濾波器以提高陣列麥克風的訊噪比與指向性。再搭配頭部轉移函數(Head Related Transfer Function)來重建空間聲場。實際的聽覺定位實驗中，利用耳機輸出之空間聲場有良好的立體空間感。此系統可應用於助聽器，不只可提高聲源的清晰度，也可清楚辨別聲源方向，更可讓聽者猶如置身立體空間聲場。

Capacitive microphones are widely used in high-quality recording because of its high sensitivity and flat frequency response. Now, it can be fabricated by Micro-Electro-Mechanical System (MEMS) technology for smaller size and better performance. Taguchi method and Genetic Algorithm (GA) are proposed to optimize the performance of sensitivity and bandwidth of the condenser microphone. Then, the parameters of diaphragm length, diaphragm thickness and air gap height can be optimized. Therefore, the MEMS condenser microphone fabrication processes are designed with the optimized parameters. Array signal processing is utilized to enhance Signal-to-Noise Ratio (SNR) and directivity of a microphone array. The comparisons between a broadside and an endfire array are illustrated. Delay-sum and superdirective methods are presented to do the Direction of Arrival (DOA) estimation and beamforming. Directivity analyses are discussed in the cases of beamwidth, directivity and SNRs. Three-dimensional (3D) spatial sound is reconstructed by inverse-filtering with Head Related Transfer Functions (HRTFs). It is implemented using TMS320C3X Digital Signal Processor (DSP) to realize 3D spatial sound. Therefore, it can be utilized for hearing aids or other applications to enhance SNR and directivity.